How to measure the ROI of handwritten notes

To measure the ROI of handwritten notes, send cards to a random share of an audience, hold back the rest as a control group, and compare what each group does next: repeat purchases, renewals, reviews, referrals. The difference between the two groups is the lift, and it is the only number worth trusting.

Be skeptical of the headline numbers

Search for handwritten note results and you will find big claims: open rates near 100%, response rates many times higher than print. We looked for the methodology behind them while researching this market and could not find any: no sample sizes, no control groups. They may be true for someone, somewhere, but you cannot plan a budget on them.

The best independent evidence is about handwriting in general, not robots. In a well-known 2005 study by Randy Garner, a handwritten sticky note asking people to complete a survey lifted responses to 76%, against 36% without one. The UK Behavioural Insights Team reports that the Irish tax authority later tested a similar handwritten note in a randomized trial and saw 36% respond against 19.2% (source). Those results are encouraging, but they are not your customers or your offer. So test it.

Set up a holdout test

Pick one moment. A first purchase, a renewal coming up, a customer milestone. One trigger per test keeps the result readable. See when to send handwritten cards for ideas.
Split at random. When the event fires, assign each person to "card" or "holdout" by a coin flip in your code, for example a hash of their ID. Never let anyone choose who gets a card; that biases the result toward your best customers.
Size the holdout. 10% to 20% held back is common. Smaller holdouts need more volume to show a clear difference.
Record the assignment. Store which group each person was in. With notecardHQ you can also put it in the note's context, which is stored with the note and never written on the card.
Send the cards. One API call per person in the card group. Use an idempotency key per person and event so a retry never sends twice; the quickstart shows how.

What to measure

Choose the outcome before you send, and measure the same thing for both groups:

Retention: renewal rate or churn over the next 90 days.
Revenue: repeat purchase rate or revenue per customer over 60 to 90 days.
Advocacy: reviews, referrals, social mentions, replies.

Then compare: if 30% of the card group bought again and 24% of the holdout did, the card added six points. Multiply that lift by what a repeat customer is worth, subtract the cost of the cards, and you have ROI you can defend.

How long to wait

Give it time. A card arrives days after the event and its effect plays out over weeks. Measure at 30 days for a first read and 90 days for the decision. Do not stop the test the day a good number appears; early results swing.

Common mistakes

Comparing to last quarter. Seasons, prices and products change. Only a concurrent holdout isolates the card.
Sending only to your best customers. They would have bought anyway, so the card looks better than it is.
Measuring opens. A card has no open tracking. Measure what people do, not whether they saw it.
Changing the copy midway. Test one note at a time. Copy tips are in how to write a handwritten note for customers.

Keep the cost side honest

Count the full cost per card: card, handwriting, envelope and postage. notecardHQ prices are all-in, from $3.25 to $1.95 per card in the US depending on monthly volume, with no monthly fee; the cost calculator works out your number and compares it with other services.

Ready to run a test? Start free with a test key, then switch to live when your holdout is wired up.